Easier/Faster way to convert Hex numbers to binary?

11 views (last 30 days)
I have an input file that has multiple columns of data, I am only interested in two of them. Right now I have both columns importing into MatLab and the first one, timestamps, is fine and I don't have any problems with it. The next one however, is a HEX number in the format 0x ####, I don't like how long it takes to run.
Currently I'm importing my dataArray,
filename = 'D:\MatLab Files\Import Files\InputFile.txt';
delimiter = ',';
startRow = 2;
formatSpec = '%*s%s%*s%s%s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%[^\n\r]';
fileID = fopen(filename,'r');
dataArray = textscan(fileID, formatSpec, 'Delimiter', delimiter, 'HeaderLines' ,startRow-1, 'ReturnOnError', false);
fclose(fileID);
dataArray{1} = datenum(dataArray{1}, 'HH:MM:SS.FFF');
Timestamp = dataArray{:, 1};
then having to run:
MsgInfo = dec2bin(hex2dec(cellfun(@(x)(x(4:end)),dataArray{:, 2}, 'UniformOutput', false)),16);
This gets me my variable into a 16 character binary number so that I can then get the bits I need out of the data. I have a total of seven bits that get set depending on if there are any errors detected in the data stream. This is how I'm currently doing this:
% using the sum function to total the number of times the bit is set to 1, giving me the total number of errors for each error type
ManchesterErrors = sum(bind2dec(MSGInfo(:,16)));
ParityErrors = sum(bind2dec(MSGInfo(:,15)));
OverRunErrors = sum(bind2dec(MSGInfo(:,14)));
TimeOutErrors = sum(bind2dec(MSGInfo(:,13)));
BroadcastMsg = sum(bind2dec(MSGInfo(:,11)));
ModeCodeNoData = sum(bind2dec(MSGInfo(:,10)));
TotalMessageErrors = sum(bind2dec(MSGInfo(:,9)));
However, this adds a lot of time (30+ minutes) to my data analysis. I'm hoping to speed this up. I found the following on the forums,
typecast(uint16(sscanf('MYHEXSTRINGHERE', '%x')), 'int16')
and tried to do this:
MSGInfoHexNumbers = cellfun(@(x)(x(4:end)),dataArray{:, 2},'UniformOutput',false); %outputs a Nx1 cell array (iscellstr = 1)
MSGInfoBinaryNumbes = typecast(uint16(sscanf(MSGInfoHexNumbers,'%x')),'int16');
but I can't get it to work for my data. I keep getting the error "First argument must be a string".
I'm using MatLab R2012b if this helps when providing answers, and I've attached a sample file (Test File.txt) of my input data, and a copy of my code to date (Hex 2 binary help.txt).

Answers (2)

Walter Roberson
Walter Roberson on 23 Dec 2015
MSGInfoBinaryNumbes = cellfun(@(S) typecast(uint16(sscanf(S(4:end),'%x')),'int16'), dataArray(:,2) );
  5 Comments
joshmartinmont
joshmartinmont on 23 Dec 2015
I timed out
cellfun(@(S) typecast(uint16(sscanf(S(4:end),'%x')),'int16'), dataArray{:, 2},'UniformOutput',false );
versus
hex2dec(cellfun(@(x)(x(4:end)),dataArray{:, 2},'UniformOutput',false))
using sscanf it took ~ 1:50 and using hex2dec it took ~ 0:50 to get the same answer of Hex to Decimal. So I guess that's not where I'm hanging up in time. Let me run some Tic/Toc and see where it's hanging up the most at and see if we can make it faster.
dpb
dpb on 23 Dec 2015
Edited: dpb on 23 Dec 2015
Try the profiler???
I'm not sure you need the typecast nor uint16 (altho they're not likely contributing much), do you? I also don't know if a format string of '0x %X' and eliminating the substring addressing would help much or not, but I'd give it a shot.

Sign in to comment.


Friedrich Welck
Friedrich Welck on 22 Mar 2016
MsgInfo = dec2bin(hex2dec(cellfun(@(x)(x(4:end)),dataArray{:, 2}, 'UniformOutput', false)),16);
is slow (cellfun is rarely fast). hex2dec, base2dec etc. work fastest on character arrays. Try
dataArray{2} = char(dataArray{2});
MsgInfo = dec2bin(base2dec(dataArray{2}(:,4:7),16),16);
Also there's no need to convert back to decimal for summation of errors. Try
ManchesterErrors = sum(MsgInfo(:,16)=='1');
ParityErrors = sum(MsgInfo(:,15)=='1');
OverRunErrors = sum(MsgInfo(:,14)=='1');
TimeOutErrors = sum(MsgInfo(:,13)=='1');
BroadcastMsg = sum(MsgInfo(:,11)=='1');
ModeCodeNoData = sum(MsgInfo(:,10)=='1');
TotalMessageErrors = sum(MsgInfo(:,9)=='1');

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!